On the recommendation of The Finnish Society for Computer Science, The Finnish Information-Processing Research Foundation has awarded its dissertation reward 2011 to Wilhelmiina Hämäläinen for her thesis ‘Efficient search for statistically significant dependency rules in binary data’ that she has completed for the Department of Computer Science at the University of Helsinki.

Statistical dependencies help us understand cause-and-effect relations, such as which genes predispose a person to certain diseases and which genes protect us from disease. Today, there is an enormous amount of data that needs analysing available in nearly all walks of life. The problem is that all dependencies cannot be studied with the usual statistical tools or computer programs. The data often contains at least hundreds or even tens of thousands of variables, and it is computationally unfeasible to study all possible dependency rules. This research has developed the necessary efficient computation methods for searching for the most significant dependency rules in binary data, where each variable may have only two values. In addition to gene research, this kind of data occurs naturally in e.g. biology (plants and animals occurring in different observation areas) and in market research (market-basket analysis, i.e. what products each customer has bought). However, if there are multi-variables in the data, they can be represented in binary form.

In comparison with earlier data mining methods, the methods developed in this research are both more efficient and reliable. The computer program developed as a result of this research can use an ordinary PC to search for the most significant dependencies in bodies of data that contain tens of thousands of variables. Wilhelmiina Hämäläinen’s contribution is thus remarkable for the statistically competent data mining of binary data.

The work was supervised by Professor Matti Nykänen from the University of Eastern Finland.

The prize was awarded at the Computer Science fair in Espoo on 30-31 May, 2011.

More information: http://www.tkts.fi/tietojenk%C3%A4sittelytieteen-v%C3%A4it%C3%B6skirjapalkinto-wilhelmiina-h%C3%A4m%C3%A4l%C3%A4iselle-tilastotieteellisesti-p%C3%A4tev%C3%A4n.